Το επόμενο κελί περιέχει όλα τα μόντουλ που χρειάζονται να εισαχθούν στη συνέχεια και ΠΡΕΠΕΙ ΝΑ ΕΚΤΑΛΕΣΘΕΙ ΑΠΟ ΤΗΝ ΑΡΧΗ.¶
In [1]:
# import math
import random
import csv
import urllib2
# from collections import Counter
import numpy as np
import pandas as pd
from pandas.tools.plotting import scatter_matrix
# from scipy import stats, integrate
import seaborn as sns # pip install seaborn
sns.set_style("white") # For seaborn to show axes in iPython Notebook
import matplotlib.pyplot as plt
from ggplot import *
# from matplotlib import cm
# For printing plots in iPython Notebook cells
%matplotlib inline
0. ΠΟΛΥΜΕΤΑΒΛΗΤΑ ΔΕΔΟΜΕΝΑ¶
0.1 Εμπερικλειόμενες (Nested) Λίστες Τυχαίων Αριθμών¶
0.1a dfrdata¶
Λίστες στηλών (εκτός από λίστα headers)¶
In [2]:
n = 21 # Number of number of rows (cases or items)
m = 5 # Number of columns (variables)
headers = ['names', 'height', 'weight', 'age']
names = ['John','Jane','Jack','Jenny','Andy','Anne','Bill','Pete','Mary','Sarah','Katy','Brian','David','Helen','Mike',
'Ruth','Alex','Keith','Linda','Dawn']
height = [x/100. for x in range(159,198,2)]
random.shuffle(height)
weight = [random.randrange(500,1000,3)/10. for _ in range(20)]
age =[random.choice(range(27,67)) for _ in range(20)]
Μετατροπή σε λίστες γραμμών (arrays)¶
In [3]:
rdata = []
for i in range(20):
rdata.append([names[i],height[i],weight[i],age[i]])
In [4]:
dfrdata = pd.DataFrame(rdata, columns=headers)
dfrdata
Out[4]:
Πρόσβαση σε λίστες στηλών/μεταβλητών από DataFrame¶
In [5]:
print dfrdata['weight'], type(dfrdata['weight'])
print dfrdata['weight'].tolist(), type(dfrdata['weight'].tolist())
print list(dfrdata['weight'])
list(dfrdata['weight']) == dfrdata['weight'].tolist()
Out[5]:
Πρόσβαση στις (μοναδικές) τιμές μιας στήλης/μεταβλητής¶
In [6]:
print dfrdata.names.unique()
print dfrdata.age.unique()
print dfrdata.names.unique().tolist()
print dfrdata.age.unique().tolist()
In [7]:
print pd.unique(dfrdata.names.ravel())
print pd.unique(dfrdata.age.ravel())
print pd.unique(dfrdata.names.ravel()).tolist()
print pd.unique(dfrdata.age.ravel()).tolist()
Σώσιμο DataFrame σε csv¶
In [8]:
dfrdata.to_csv('data1.csv', sep='\t', index=False) # sep = ';','\t'
# dfrdata.to_csv('data1.csv', sep='\t', encoding='utf-8')
Άνοιγμα/εισαγωγή csv ως DataFrame¶
In [9]:
dfrdata1 = pd.read_csv('data1.csv', sep='\t')
dfrdata1
Out[9]:
0.2 Πολυμεταβλητά Δεδομένα ως DataFrames¶
Κατασκευή DataFrame με Τυχαίους Αριθμούς¶
0.2a dfrdata2¶
In [10]:
dfrdata2 = pd.DataFrame(np.random.randn(10,6))
print dfrdata2.columns
dfrdata2
Out[10]:
In [11]:
dfrdata2 = dfrdata2.rename(columns={0:'Var0',1:'Var1',2:'Var2',3:'Var3',4:'Var4',5:'Var5',})
dfrdata2
Out[11]:
In [12]:
print dfrdata2['Var2'], type(dfrdata2['Var2'])
print dfrdata2['Var2'].tolist(), type(dfrdata2['Var2'].tolist())
print list(dfrdata2['Var2'])
list(dfrdata2['Var2']) == dfrdata2['Var2'].tolist()
Out[12]:
In [13]:
dfrdata2_vars = []
for i in dfrdata2.columns:
dfrdata2_vars.append(list(dfrdata2[i]))
# dfrdata2_vars
Κατασκευή DataFrame από λεξικό¶
0.2b ddf¶
In [14]:
datadict = {'occupation': ['painter','actor','author','singer','student','manager','singer'],
'city': ['Athens','Paris','London','Rome','Tokyo','New York','Lima'],
'year': [1980,1998,2003,1977,1992,1977,1985],
'reports': [5,14,22,8,31,26,17]}
for key, value in datadict.items():
print key, 'has length', len(value)
ddf = pd.DataFrame(datadict, index = range(7))
print ddf.shape
ddf
Out[14]:
In [15]:
print ddf.city.unique()
print ddf.occupation.unique()
print ddf.reports.unique()
print ddf.year.unique()
# print pd.unique(ddf.city.ravel())
# print pd.unique(ddf.occupation.ravel())
# print pd.unique(ddf.reports.ravel())
# print pd.unique(ddf.year.ravel())
0.2c mtcars¶
In [16]:
mtcars = pd.read_csv("https://vincentarelbundock.github.io/Rdatasets/csv/datasets/mtcars.csv")
print mtcars.shape
print mtcars.columns
mtcars
Out[16]:
In [17]:
mtcars = mtcars.rename(columns={'Unnamed: 0': 'Cars'})
mtcars.columns
Out[17]:
0.2d diamonds¶
In [18]:
diamonds = pd.read_csv("https://vincentarelbundock.github.io/Rdatasets/csv/ggplot2/diamonds.csv")
print diamonds.shape
print diamonds.info()
diamonds = diamonds.rename(columns={'Unnamed: 0': '#'})
print diamonds.columns
diamonds.isnull().any().any() # Check for missing values
Out[18]:
In [19]:
diamonds.head(5)
# diamonds.head(n=5)
Out[19]:
In [20]:
diamonds.iloc[range(5)]
Out[20]:
In [21]:
diamonds.ix[range(5)]
Out[21]:
In [22]:
diamonds.iloc[range(1000,1005)]
Out[22]:
In [23]:
diamonds.ix[range(1000,1005)]
Out[23]:
In [24]:
diamonds.iloc[range(53940)[-5:]]
Out[24]:
In [25]:
diamonds.ix[range(53940)[-5:]]
Out[25]:
In [26]:
diamonds.tail(5)
Out[26]:
In [27]:
print diamonds.cut.unique()
print diamonds.color.unique()
print diamonds.clarity.unique()
# print pd.unique(diamonds.cut.ravel())
# print pd.unique(diamonds.color.ravel())
# print pd.unique(diamonds.clarity.ravel())
0.2e anscombe¶
In [28]:
anscombe = pd.read_csv("https://vincentarelbundock.github.io/Rdatasets/csv/datasets/anscombe.csv")
print anscombe.shape
print anscombe.columns
anscombe
Out[28]:
tips¶
In [29]:
tips = pd.read_csv("https://vincentarelbundock.github.io/Rdatasets/csv/reshape2/tips.csv")
print tips.shape
print tips.columns
tips.head(n=4)
Out[29]:
In [30]:
print tips.sex.unique()
print tips.smoker.unique()
print tips.day.unique()
print tips.time.unique()
# print pd.unique(tips.sex.ravel())
# print pd.unique(tips.smoker.ravel())
# print pd.unique(tips.day.ravel())
# print pd.unique(tips.time.ravel())
0.3 Χειρισμός Ελλειπουσών Τιμών¶
Κατασκευή DataFrame με Τυχαίους Αριθμούς και Ελλείπουσες Τιμές¶
0.3a dfrn¶
In [31]:
dfrn = pd.DataFrame(np.random.randn(10,4))
dfrn.iloc[1:4,0] = np.nan
dfrn.iloc[5,2] = np.nan
dfrn.iloc[7:9,3] = np.nan
print dfrn.shape
dfrn
Out[31]:
In [32]:
dfrn = pd.DataFrame(np.random.randn(10,4))
dfrn.iloc[1:4,0] = None
dfrn.iloc[5,2] = None
dfrn.iloc[7:9,3] = None
print dfrn.shape
dfrn
Out[32]:
In [33]:
print dfrn.isnull().any().any()
print dfrn.isnull().any()
print dfrn.isnull().sum().sum()
dfrn.isnull().sum()
Out[33]:
In [34]:
dfrn = dfrn.dropna()
print dfrn.isnull().any().any()
dfrn.isnull().sum().sum()
print dfrn.shape
dfrn
Out[34]:
0.3b airquality¶
In [35]:
airquality = pd.read_csv("https://vincentarelbundock.github.io/Rdatasets/csv/datasets/airquality.csv")
print airquality.shape
print airquality.columns
# airquality
In [36]:
print airquality.isnull().any().any()
print airquality.isnull().any()
print airquality.isnull().sum().sum()
airquality.isnull().sum()
Out[36]:
In [37]:
airquality = airquality.dropna()
print airquality.isnull().any().any()
print airquality.isnull().sum().sum()
print airquality.shape
print airquality.columns
airquality.head(n=4)
Out[37]:
0.3c brain_size¶
In [38]:
brainsize = pd.read_csv("brain_size.csv",sep=';')
print brainsize.shape
print brainsize.isnull().any().any()
print brainsize.isnull().any()
print brainsize.isnull().sum().sum()
brainsize.isnull().sum()
Out[38]:
In [39]:
brainsize
Out[39]:
In [40]:
df = brainsize
df['Weight'].replace('.', np.nan, inplace=True)
df['Height'].replace('.', np.nan, inplace=True)
brainsize = df.dropna()
print brainsize.isnull().any().any()
print brainsize.isnull().sum().sum()
print brainsize.shape
brainsize
Out[40]:
1. ΒΑΣΙΚΕΣ ΠΟΛΥΜΕΤΑΒΛΗΤΕΣ ΣΤΑΤΙΣΤΙΚΕΣ¶
1.1 Συνδιακύμανση¶
Η συνδιακύμανση δυο μεταβλητών $x = (x_1, x_2, \ldots, x_n)$ και $y = (y_1, y_2, \ldots, y_n)$ με μέσες τιμές $\bar{x}$ και $\bar{y}$, αντιστοίχως, ορίζεται ως
\begin{equation} \frac{1}{n} \sum_{i = 1}^n (x_i - \bar{x}) (y_i - \bar{y}) \end{equation}Η δειγματική συνδιακύμανση των μεταβλητών αυτών ορίζεται ως
\begin{equation} \frac{1}{n - 1} \sum_{i = 1}^n (x_i - \bar{x}) (y_i - \bar{y}) \end{equation}In [41]:
def covariance(x,y):
return sum([(xi - np.mean(x))*(yi - np.mean(y)) for (xi, yi) in zip(x,y)])/float(len(x))
def scovariance(x,y):
return sum([(xi - np.mean(x))*(yi - np.mean(y)) for (xi, yi) in zip(x,y)])/float(len(x)-1)
1.1a dfrdata2¶
In [42]:
x = dfrdata2_vars[1]
y = dfrdata2_vars[3]
print covariance(x,y), scovariance(x,y)
print np.cov(x,y)[1,0]
print np.cov(x,y)
dfrdata2.cov()
Out[42]:
1.1b mtcars¶
In [43]:
print mtcars[['hp','wt']].cov()
print 'cov(hp,wt) =', mtcars[['hp','wt']].cov().hp[1]
mtcars.cov()
Out[43]:
1.1c diamonds¶
In [44]:
print diamonds[['table','y']].cov()
print 'cov(y,table) =', diamonds[['table','y']].cov().y[0]
diamonds.cov()
Out[44]:
1.2 Συσχέτιση (Συντελεστής Συσχέτισης του Pearson)¶
Η συσχέτιση (δειγματική συσχέτιση) δυο μεταβλητών ορίζεται ως η (δειγματική) συνδιακύμανσή τους διαιρούμενη με το γινόμενο των (δειγματικών) τυπικών αποκλίσεών τους.
1.2a dfrdata2¶
In [45]:
print dfrdata2[['Var1','Var4']].corr()
print 'corr(Var1,Var4) =', dfrdata2[['Var1','Var4']].corr().Var1[1]
print 'corr(Var1,Var4) =', dfrdata2[['Var1','Var4']].corr().Var1[0]
dfrdata2.corr()
Out[45]:
1.2b mtcars¶
In [46]:
print mtcars[['hp','wt']].corr()
print 'corr(hp,wt) =', mtcars[['hp','wt']].corr().wt[0]
mtcars.corr()
Out[46]:
1.2c diamonds¶
In [47]:
print diamonds[['table','y']].corr()
print 'corr(table,y) =', diamonds[['table','y']].corr().table[1]
diamonds.corr()
Out[47]:
1.3 Πίνακες Συμπτώσεων¶
1.3a tips¶
In [48]:
tips.info()
In [49]:
# Find categorical variables (columns):
tips.applymap(lambda x: isinstance(x, str)).any()
Out[49]:
In [50]:
cv_tips = ['sex','smoker','day','time']
print cv_tips
ld_tips = {}
for i in tips.columns.tolist():
ld_tips[i] = len(tips[i].unique())
print ld_tips
In [51]:
pd.crosstab(tips['size'], tips.day)
Out[51]:
In [52]:
pd.crosstab(tips['size'], [tips.sex, tips.smoker, tips.time])
Out[52]:
In [53]:
sizedf = pd.crosstab(tips['size'], tips.day)
sexdf = pd.crosstab(tips['size'], tips.sex)
smokerdf = pd.crosstab(tips['size'], tips.smoker)
timedf = pd.crosstab(tips['size'], tips.time)
sizedf.join([sexdf,smokerdf,timedf])
Out[53]:
In [54]:
pd.crosstab(tips['size'], tips.tip)
Out[54]:
1.3b mtcars¶
In [55]:
mtcars.info()
In [56]:
# Find categorical variables (columns):
mtcars.applymap(lambda x: isinstance(x, str)).any()
Out[56]:
In [57]:
cv_mtcars = ['Cars']
ld_mtcars = {}
for i in mtcars.columns.tolist():
ld_mtcars[i] = len(mtcars[i].unique())
print ld_mtcars
In [58]:
pd.crosstab(mtcars.Cars, mtcars.carb).head(5)
Out[58]:
In [59]:
pd.crosstab(mtcars.Cars, [mtcars.cyl, mtcars.vs, mtcars.am, mtcars.gear]).head(5)
Out[59]:
In [60]:
import copy
mtcars1 = copy.deepcopy(mtcars)
mtcars1['cyl'].replace(to_replace=mtcars1['cyl'].unique().tolist(),value=['cyl4','cyl6','cyl8'],inplace=True)
mtcars1['vs'].replace(to_replace=mtcars1['vs'].unique().tolist(),value=['vs0','vs1'],inplace=True)
mtcars1['am'].replace(to_replace=mtcars1['am'].unique().tolist(),value=['am0','am1'],inplace=True)
mtcars1['gear'].replace(to_replace=mtcars1['gear'].unique().tolist(),value=['gear3','gear4','gear5'],inplace=True)
mtcars1['carb'].replace(to_replace=mtcars1['carb'].unique().tolist(),value=['carb1','carb2','carb3','carb4','carb6','carb8'],inplace=True)
carbdf = pd.crosstab(mtcars1['Cars'], mtcars1.carb)
vsdf = pd.crosstab(mtcars1['Cars'], mtcars1.vs)
amdf = pd.crosstab(mtcars1['Cars'], mtcars1.am)
cyldf = pd.crosstab(mtcars1['Cars'], mtcars1.cyl)
carbdf.join([vsdf,amdf,cyldf])
Out[60]:
1.3c diamonds¶
In [61]:
diamonds.info()
In [62]:
# Find categorical variables (columns):
diamonds.applymap(lambda x: isinstance(x, str)).any()
Out[62]:
In [63]:
cv_diamonds = ['cut','color','clarity']
ld_diamonds = {}
for i in diamonds.columns.tolist():
ld_diamonds[i] = len(diamonds[i].unique())
print ld_diamonds
In [64]:
pd.crosstab(diamonds.price, diamonds.clarity).head(5)
Out[64]:
In [65]:
pd.crosstab(diamonds.price, diamonds.cut).head(5)
Out[65]:
In [66]:
pd.crosstab(diamonds.price, diamonds.color).head(5)
Out[66]:
In [67]:
cutdf = pd.crosstab(diamonds.price, diamonds.cut)
colordf = pd.crosstab(diamonds.price, diamonds.color)
claritydf = pd.crosstab(diamonds.price, diamonds.clarity)
cutdf.join([colordf,claritydf]).head(5)
Out[67]:
1.4 Στατιστικές Συμπτώσεων¶
1.4a tips¶
In [68]:
cv_tips
Out[68]:
In [69]:
tips.groupby('sex').describe()
Out[69]:
In [70]:
tips.groupby('sex').mean()
Out[70]:
In [71]:
tips['total_bill'].groupby(tips['time']).describe()
Out[71]:
In [72]:
tips['total_bill'].groupby(tips['time']).describe().unstack()
Out[72]:
In [73]:
tips['total_bill'].groupby(tips['time']).std()
Out[73]:
1.4b mtcars¶
In [74]:
cv_mtcars
Out[74]:
In [75]:
mtcars.groupby('Cars').describe().head(8)
Out[75]:
In [76]:
mtcars.groupby('Cars').max().head(3)
Out[76]:
In [77]:
mtcars['mpg'].groupby(mtcars['qsec']).describe().unstack().head(3)
Out[77]:
In [78]:
mtcars['mpg'].groupby(mtcars['qsec']).max().head(3)
Out[78]:
1.4c diamonds¶
In [79]:
cv_diamonds
Out[79]:
In [80]:
diamonds.groupby('color').describe().head(8)
Out[80]:
In [81]:
diamonds.groupby('color').min().head(3)
Out[81]:
In [82]:
diamonds['price'].groupby(diamonds['color']).describe().unstack().head(3)
Out[82]:
In [83]:
diamonds['price'].groupby(diamonds['clarity']).mean().head(3)
Out[83]:
2. ΣΧΕΔΙΑΓΡΑΜΜΑΤΑ ΠΟΛΥΜΕΤΑΒΛΗΤΩΝ ΣΤΑΤΙΣΤΙΚΩΝ¶
2.1 Σχεδιαγράμματα Πολλαπλών Μεταβλητών¶
2.1a tips¶
In [84]:
# Find numerical variables
print set(tips.columns).difference(set(cv_tips))
In [85]:
tips.drop('Unnamed: 0',inplace=True,axis=1)
In [86]:
tips.plot(figsize=(10,10))
Out[86]:
In [87]:
tips.plot(kind="area", figsize=(10,10), stacked=False)
Out[87]:
In [88]:
tips.plot(kind="area", figsize=(10,10), stacked=True)
Out[88]:
In [89]:
ax = tips.plot(kind="bar", figsize=(10,10), stacked=False)
ax.xaxis.set_major_formatter(plt.NullFormatter())
In [90]:
ax = tips.plot(kind="bar", figsize=(10,10), stacked=True)
ax.xaxis.set_major_formatter(plt.NullFormatter())
In [91]:
tips.plot(kind="hist", figsize=(10,10), stacked=True)
Out[91]:
In [92]:
tips.plot(kind="kde", xlim=(-3,63), figsize=(10,10))
Out[92]:
In [93]:
tips['total_bill'].plot.kde(xlim=(-3,63), figsize=(10,10))
Out[93]:
2.1b mtcars¶
In [94]:
# Find numerical variables
cv_mtcars = ['Cars']
print set(mtcars.columns).difference(set(cv_mtcars))
In [95]:
mtcars.plot(figsize=(10,10))
Out[95]:
In [96]:
mtcars.plot(kind="area", figsize=(10,10), stacked=False)
Out[96]:
In [97]:
mtcars.plot(kind="area", figsize=(10,10), stacked=True)
Out[97]:
In [98]:
mtcars.plot(kind="bar", figsize=(10,10), stacked=False)
Out[98]:
In [99]:
mtcars.plot(kind="bar", figsize=(10,10), stacked=True)
Out[99]:
In [100]:
mtcars.plot(kind="hist", figsize=(10,10), stacked=True)
Out[100]:
In [101]:
# mtcars.describe()
In [102]:
# mtcars.plot(kind="kde", figsize=(10,10))
mtcars.plot(kind="kde", xlim=(-50,501), figsize=(10,10))
Out[102]:
In [103]:
mtcars[['vs','am']].plot.kde(figsize=(10,10))
Out[103]:
2.1c diamonds¶
In [104]:
# Find numerical variables
print set(diamonds.columns).difference(set(cv_diamonds))
In [105]:
diamonds.drop('#',inplace=True,axis=1)
In [106]:
diamonds.plot(figsize=(10,10))
Out[106]:
In [107]:
diamonds.describe()
Out[107]:
In [108]:
diamonds.plot(kind="area", figsize=(10,10), stacked=False)
Out[108]:
In [109]:
diamonds.plot(kind="area", figsize=(10,10), stacked=True)
Out[109]:
In [110]:
diamonds_sample = diamonds.sample(100)
In [111]:
ax = diamonds_sample.plot(kind="bar", figsize=(10,10), stacked=False)
ax.xaxis.set_major_formatter(plt.NullFormatter())
In [112]:
ax = diamonds_sample.plot(kind="bar", figsize=(10,10), stacked=True)
ax.xaxis.set_major_formatter(plt.NullFormatter())
In [113]:
diamonds_sample1 = diamonds[['carat','depth','table','x','y','z']].sample(100)
ax = diamonds_sample1.plot(kind="bar", figsize=(10,10), stacked=True)
ax.xaxis.set_major_formatter(plt.NullFormatter())
In [114]:
diamonds_sample.plot(kind="hist", figsize=(10,10), stacked=True)
Out[114]:
In [115]:
diamonds_sample1.describe()
Out[115]:
In [116]:
diamonds_sample1.plot(kind="kde", figsize=(10,10))
# diamonds_sample1.plot(kind="kde", xlim=(-50,501), figsize=(10,10))
Out[116]:
In [117]:
diamonds_sample1[['carat','z']].plot.kde(figsize=(10,10))
Out[117]:
2.2 Σχεδιαγράμματα Συμπτώσεων Δυο Μεταβλητών¶
2.2a tips¶
In [118]:
print ld_tips
cv_tips
Out[118]:
In [119]:
pd.crosstab(tips.day, tips.time)
Out[119]:
In [120]:
plt.figure(figsize=(5,7))
sns.heatmap(pd.crosstab(tips.day, tips.time))
Out[120]:
In [121]:
pd.crosstab(tips.day, tips.time).plot(kind="area", figsize=(10,10), stacked=False)
Out[121]:
In [122]:
pd.crosstab(tips.day, tips.time).plot(kind="area", figsize=(10,10), stacked=True)
Out[122]:
In [123]:
pd.crosstab(tips.day, tips.time).plot(kind="bar", figsize=(10,10), stacked=False)
Out[123]:
In [124]:
pd.crosstab(tips.day, tips.time).plot(kind="bar", figsize=(10,10), stacked=True)
Out[124]:
In [125]:
pd.crosstab(tips.day, tips.time).plot(kind="hist", figsize=(10,10), stacked=True)
Out[125]:
2.2b mtcars¶
In [126]:
print ld_mtcars
cv_mtcars
Out[126]:
In [127]:
pd.crosstab(mtcars.cyl,mtcars.gear)
Out[127]:
In [128]:
plt.figure(figsize=(6,5))
sns.heatmap(pd.crosstab(mtcars.cyl,mtcars.gear))
Out[128]:
In [129]:
pd.crosstab(mtcars.cyl,mtcars.gear).plot(kind="area", figsize=(10,10), stacked=False)
Out[129]:
In [130]:
pd.crosstab(mtcars.cyl,mtcars.gear).plot(kind="area", figsize=(10,10), stacked=True)
Out[130]:
In [131]:
pd.crosstab(mtcars.cyl,mtcars.gear).plot(kind="bar", figsize=(10,10), stacked=False)
Out[131]:
In [132]:
pd.crosstab(mtcars.cyl,mtcars.gear).plot(kind="bar", figsize=(10,10), stacked=True)
Out[132]:
In [133]:
pd.crosstab(mtcars.cyl,mtcars.gear).plot(kind="hist", figsize=(10,10), stacked=True)
Out[133]:
2.2c diamonds¶
In [134]:
print ld_diamonds
cv_diamonds
Out[134]:
In [135]:
pd.crosstab(diamonds.clarity,diamonds.color)
Out[135]:
In [136]:
plt.figure(figsize=(12,10))
sns.heatmap(pd.crosstab(diamonds.clarity,diamonds.color))
Out[136]:
In [137]:
pd.crosstab(diamonds.clarity,diamonds.color).plot(kind="area", figsize=(10,10), stacked=False)
Out[137]:
In [138]:
pd.crosstab(diamonds.clarity,diamonds.color).plot(kind="area", figsize=(10,10), stacked=True)
Out[138]:
In [139]:
pd.crosstab(diamonds.clarity,diamonds.color).plot(kind="bar", figsize=(10,10), stacked=False)
Out[139]:
In [140]:
pd.crosstab(diamonds.clarity,diamonds.color).plot(kind="bar", figsize=(10,10), stacked=True)
Out[140]:
In [141]:
pd.crosstab(diamonds.clarity,diamonds.color).plot(kind="hist", figsize=(10,10), stacked=True)
Out[141]:
2.3 Διαγράμματα Διασποράς¶
2.3a df¶
In [142]:
df = pd.DataFrame(np.random.randn(5000, 2), columns=['a', 'b'])
df['b'] = df['b'] + np.arange(5000)
df.plot.scatter(x='a', y='b', figsize=(10,10))
Out[142]:
In [143]:
sns.jointplot(x="a", y="b", data=df, space=0, size=10)
Out[143]:
In [144]:
df.plot.hexbin(x='a', y='b', gridsize=30, figsize=(12,10))
Out[144]:
In [145]:
sns.jointplot("a", "b", data=df, kind="hex", gridsize=30, space=0, size=10)
Out[145]:
In [146]:
sns.jointplot("a", "b", data=df, kind="kde", space=0, size=10)
Out[146]:
2.3b tips¶
In [147]:
# Find numerical variables
print set(tips.columns).difference(set(cv_tips))
ld_tips
Out[147]:
In [148]:
for i in tips.columns:
print i, len(tips[i].unique())
tips.head(5)
Out[148]:
In [149]:
tips.plot.scatter(x='total_bill',y='tip',figsize=(10,10))
Out[149]:
In [150]:
sns.jointplot(x="total_bill", y="tip", data=tips, space=0, size=10)
Out[150]:
In [151]:
tips.plot.hexbin(x='total_bill', y='tip', gridsize=25, figsize=(12,10))
Out[151]:
In [152]:
sns.jointplot("total_bill", "tip", data=tips, kind="hex", gridsize=25, space=0, size=10)
Out[152]:
In [153]:
sss=scatter_matrix(tips[['total_bill','tip','size']], alpha=0.9, figsize=(12, 12), color='black', diagonal='hist')
In [154]:
sns.pairplot(tips[['total_bill','tip','size']],kind='scatter',diag_kind='hist',size=3.5)
Out[154]:
In [155]:
sns.pairplot(tips, hue="day", size=3.5)
Out[155]:
2.3c mtcars¶
In [156]:
# Find numerical variables
print set(mtcars.columns).difference(set(cv_mtcars))
ld_mtcars
Out[156]:
In [157]:
mtcars.plot.scatter(x='wt',y='qsec',figsize=(10,10))
Out[157]:
In [158]:
ax=mtcars.plot.scatter(x='mpg',y='hp',label='hp per mpg',figsize=(10,10),color='blue')
mtcars.plot.scatter(x='mpg',y='disp',label='disp per mpg',figsize=(10,10),ax=ax,color='red')
Out[158]:
In [159]:
ax=mtcars.plot.scatter(x='mpg',y='qsec',label='qsec per mpg',figsize=(10,10),color='green')
mtcars.plot.scatter(x='mpg',y='wt',label='wt per mpg',figsize=(10,10),ax=ax,color='black')
Out[159]:
In [160]:
mtcars.plot.scatter(x='mpg',y='hp', figsize=(10,10))
Out[160]:
In [161]:
sns.jointplot(x="mpg", y="hp", data=mtcars, space=0, size=10)
Out[161]:
In [162]:
mtcars.plot.hexbin(x='mpg', y='hp', gridsize=25, figsize=(12,10))
Out[162]:
In [163]:
sns.jointplot("mpg", "hp", data=mtcars, kind="hex", gridsize=25, space=0, size=10)
Out[163]:
In [164]:
sns.jointplot("mpg", "hp", data=mtcars, kind="kde", space=0, size=10)
Out[164]:
In [165]:
mtcars.plot.scatter(x='mpg', y='hp', c='qsec', s=50, figsize=(12,10))
Out[165]:
In [166]:
mtcars.plot.scatter(x='mpg', y='hp', s=mtcars['qsec']*20, figsize=(10,10))
Out[166]:
In [167]:
mtcars.plot.hexbin(x='mpg', y='hp', C='qsec', reduce_C_function=np.max, gridsize=25, figsize=(12,10))
Out[167]:
In [168]:
mtcars.plot.hexbin(x='drat', y='wt', C='qsec', reduce_C_function=np.max, gridsize=25, figsize=(12,10))
Out[168]:
In [169]:
sss= scatter_matrix(mtcars[['mpg','hp','carb','disp']], alpha=0.9, figsize=(12, 12), color='black', diagonal='kde')
In [170]:
sns.pairplot(mtcars[['mpg','hp','carb','disp']],kind='scatter',diag_kind='kde')
Out[170]:
In [171]:
mtcars1 = mtcars[['mpg','hp','carb','disp','qsec']]
sns.pairplot(mtcars1, hue="qsec")
Out[171]:
2.3d diamonds¶
In [172]:
# Find numerical variables
print set(diamonds.columns).difference(set(cv_diamonds))
ld_diamonds
Out[172]:
In [173]:
diamonds.plot.scatter(x='price', y='carat', figsize=(10,10))
Out[173]:
In [174]:
sns.jointplot(x="price", y="carat", data=diamonds, space=0, size=10)
Out[174]:
In [175]:
diamonds.plot.hexbin(x='price', y='carat', gridsize=50, figsize=(12,10))
Out[175]:
In [176]:
ax=diamonds.plot.hexbin(x='price', y='carat', gridsize=50, figsize=(12,10))
ax.set_xlim(0,4000)
ax.set_ylim(0,1)
Out[176]:
In [177]:
sns.jointplot("price", "carat", data=diamonds, kind="hex", gridsize=50, space=0, size=10)
Out[177]:
In [178]:
g=sns.jointplot("price", "carat", data=diamonds, kind="hex", gridsize=50, space=0, size=10)
g.ax_joint.set(xlim=(0, 4000), ylim=(0, 1))
Out[178]:
In [179]:
sns.jointplot("price", "carat", data=diamonds, kind="kde", space=0, size=10)
Out[179]:
In [180]:
g=sns.jointplot("price", "carat", data=diamonds, kind="kde", space=0, size=10)
g.ax_joint.set(xlim=(0, 4000), ylim=(0, 1))
Out[180]:
In [181]:
p = ggplot(aes(x='price', y='carat',color="clarity"), data=diamonds)
p + geom_point()
Out[181]:
In [182]:
p = ggplot(aes(x='price', y='carat',color="cut"), data=diamonds)
p + geom_point()
Out[182]:
In [183]:
sss = scatter_matrix(diamonds[['carat','depth','table','price']], alpha=0.9, figsize=(12, 12), color='black', diagonal='hist')
In [184]:
sns.pairplot(diamonds[['carat','depth','table','price']],kind='scatter',diag_kind='hist',size=2.8)
Out[184]:
In [185]:
diamonds1 = diamonds[['carat','depth','table','price','clarity']]
sns.pairplot(diamonds1, hue="clarity", size=2.8)
Out[185]:
2.3e iris¶
In [186]:
iris = sns.load_dataset("iris")
iris.shape
Out[186]:
In [187]:
iris.info()
In [188]:
# Find categorical/numerical variables (columns):
print iris.applymap(lambda x: isinstance(x, str)).any()
In [189]:
cv_iris = ['species']
print 'Numerical variables:', list(set(iris.columns).difference(set(cv_iris)))
ld_iris = {}
for i in iris.columns.tolist():
ld_iris[i] = len(iris[i].unique())
print ld_iris
In [190]:
g = sns.pairplot(iris)
In [191]:
g = sns.pairplot(iris, hue="species")
In [192]:
g = sns.pairplot(iris, hue="species", palette="husl")
In [193]:
g = sns.pairplot(iris, hue="species", markers=["o", "s", "D"])
In [ ]: